Tradingview MCP is a free, MIT-licensed Python server that connects AI assistants to market analysis tools. It is aimed at traders, analysts and developers who want a language model to run stock screeners, multi-timeframe technical checks and walk-forward backtests without switching between charting tabs and spreadsheets. The program runs on Windows through a ready-made build or installer, and it is also available for macOS, Linux and Docker.
Its scope goes well beyond crypto. The server covers stocks, crypto, forex and futures across global exchanges, so the same workflow can compare a Bitcoin setup with an equity or a currency pair. The current release is v0.8.0, and the project received a code update in late October 2026, which means it is actively developed. Treat the output as research material rather than a recommendation, and read the sections below before connecting it to your assistant.
What Is an MCP Server for Trading Analysis?
The Model Context Protocol, usually shortened to MCP, is an open standard that lets AI assistants call external tools and data sources during a conversation. An MCP server is the small program that exposes those tools. Tradingview MCP is one such server, and its tools cover screeners, technical analysis and backtests.
The difference from a standard chatbot is practical. A language model on its own can describe what a moving average crossover is, but it cannot fetch today's prices or run a test on a year of hourly candles. With the server connected, the assistant can request fresh screener results, calculate indicators across several timeframes and return a backtest table that you can inspect.
An example makes this concrete. You might ask the assistant to scan a watchlist for symbols pulling back to a trend line on the daily chart, then check whether the hourly chart confirms the same setup. The assistant calls the screener, runs the analysis on each candidate and summarizes the results. You still decide what to do with that information.
Who Is Tradingview MCP For?
The most natural users are discretionary traders who already work with indicators and want a structured second look at their ideas. A trader who has a rule in mind, such as buying pullbacks above a long-term average, can ask the assistant to test that rule on several symbols and report where it held up.
Developers form a second group. Because the code is MIT-licensed, it can be read, adapted and built into internal workflows, such as a daily report that runs screeners and saves the results. Students and researchers can use the backtesting tools to see how a strategy behaves across different market regimes without writing an engine from scratch.
The tool is less suited to people who want a simple point-and-click experience. Setting up an MCP server involves editing configuration files and handling paths, and the workflow assumes you are comfortable reading backtest output. Beginners may find it easier to learn the basics of indicators first and then add an assistant later.
How Walk-Forward Testing and Overfitting Work
Most backtests suffer from a simple problem. If you adjust a strategy's parameters until it performs well on past data, the result may describe that history rather than a repeatable edge. This is called overfitting or curve-fitting, and it is one of the most common reasons backtests look better than live results.
Walk-forward testing addresses the problem by separating the data. The method optimizes a strategy on an early window, which is the training period, and then measures it on the following window, which is the test period. The process repeats as the window moves forward, so each judgment is based on data that was not used to choose the parameters.
A good result from this kind of test is encouraging but not conclusive. Markets change, sample sizes can be small and a strategy that survives one period may fail in the next. Use the verdict as a filter that flags weak ideas early, then check the trade log and equity curve to understand where the results come from.
Key features
Nine backtest strategies
Resilient screener data
Rebuilt financial news feed
Walk-forward backtesting with an overfitting verdict
Hourly timeframe support
Full trade logs and equity curves
Supported coins
The documentation names Bitcoin (BTC) and Ethereum (ETH) as examples of supported coins. The server's wider scope covers stocks, crypto, forex and futures across global exchanges, so crypto is one of several asset classes rather than the only focus. Check the symbol format your market requires before running a screener.
How to install Tradingview MCP on Windows
- 1
Download the ready-made build
Download only from the official website or the project's GitHub repository at github.com/atilaahmettaner/tradingview-mcp, choosing the release marked v0.8.0 when it is listed. If the release page publishes a SHA-256 checksum or a signature, verify the file against it before running anything. Avoid mirrors and files shared through forums or chat groups.
- 2
Confirm the uvx launcher
The server is started with uvx, a command-line runner that comes with the uv tool. Open Command Prompt or PowerShell and confirm that uvx responds. On Windows it is typically located at %USERPROFILE%\.local\bin\uvx.exe, so note the full path for the configuration step.
- 3
Register the server with your AI client
Add the server to your assistant's MCP configuration. If the client does not search your PATH, use the full path to uvx. Save the file and restart the client so it loads the new entry. Path spelling is easy to get wrong, so double-check it before troubleshooting anything else.
- 4
Add your AI provider key
Enter the API key for the model provider you already use. OpenRouter is not required, so any provider you have access to will work. Store the key in the client's settings or an environment variable, and never paste it into a chat window.
-
Test the connection
Ask the assistant to run a simple screener on a liquid symbol. A returned table means the server is reachable and the market data path is working. If the request fails, look at the error message first, then review your configuration path and provider key.
System requirements
- Windows, with the ready-made build or installer
- uvx command-line launcher from the uv Python tooling, available at the full path in your configuration
- An API key from the AI provider you choose; OpenRouter is not required
- An internet connection for live market data
How to use Tradingview MCP
Once the server is connected to your AI client, you control it through plain-language requests. The tools accept parameters such as the symbol, the timeframe and the test period, so it helps to plan a request before typing it.
- Ask for a screener first to narrow a watchlist to symbols that meet your conditions.
- Run multi-timeframe analysis on a shortlisted symbol to see whether the daily and hourly views agree.
- Choose a strategy and run a backtest with the period set to '1y' or '2y'. Shorter periods do not give the SMA200 trend filter enough bars to finish its warmup.
- Compare several strategies with the comparison tool and note which ones stay consistent across timeframes.
- Run a walk-forward test on the strongest candidate and read the overfitting verdict before drawing conclusions.
- Ask for the trade log and equity curve when you want to inspect individual trades or drawdowns.
Keep requests specific. Name the market, the asset and the timeframe, and ask the assistant to explain the numbers in its own words so you understand the assumptions behind them.
Pros & cons
- Free and MIT-licensed, so the code can be read, modified and reused
- Covers stocks, crypto, forex and futures within one workflow
- Nine backtest strategies with side-by-side comparison
- Walk-forward testing with a clear overfitting verdict
- Runs on Windows, macOS, Linux and Docker
- Hourly timeframe support for intraday tests
- Requires comfort with command-line setup and editing client configuration
- Backtests need a period of one or two years to run as intended
- Documentation names only Bitcoin and Ethereum as examples, so confirm coverage for other assets
- Depends on an AI provider key that you supply yourself
Is Tradingview MCP safe?
Obtain Tradingview MCP only from the official documentation at tradingview.com/mcp/docs or from the project's GitHub repository. Check the SHA-256 checksum or signature when the release publishes one, and review the repository before running code from an unfamiliar fork. Never paste API keys or account credentials into third-party websites or chat tools you do not control. The server produces analysis, not financial advice, and it cannot promise profits. Backtests describe past data only. Before updating, check the latest version on the official website.
- Official source only. Download from github.com/atilaahmettaner/tradingview-mcp. Ignore ads and lookalike domains.
- Check the SHA-256. Compare the installer hash with the value the developer publishes before running it.
- Never share your seed phrase. No developer, support agent or website — including CryptApps — will ever ask for it.
Alternatives
Tradingview MCP is a server for AI assistants, while most programs in this group are standalone applications. Quantify is a Python trading terminal with multi-strategy backtesting and a real-time, signal-only decision desk, so it suits people who want a dedicated screen for decisions. Tradingview MCP keeps the interface in your chat window and leaves the reasoning to whichever assistant you connect.
Hummingbot focuses on market making and arbitrage, a different goal from analysis-first research. OctoBot provides a trading bot with a web interface, which is better for running automated strategies from a dashboard. StockSharp is a development platform for building trading robots in C#, which suits programmers who want full control over the code. Tradingview MCP asks less of you on the coding side but offers fewer options for building custom automation.
Specifications
- Price
- Free
- Type
- Open source
- License
- MIT
- Latest version
- v0.8.0
- Last update
- 29 Oct 2026
- Windows support
- Yes (Windows build or installer)
- Source code
- github.com/atilaahmettaner/tradingview-mcp
Frequently asked questions
Yes. The program is free and released under the MIT license, which allows you to use, copy, modify and distribute the code. Costs for the AI model you connect are separate, because you bring your own provider key. OpenRouter is not required, so you can use whichever provider you already have access to.
Yes. A Windows build or installer is available, and the server also runs on macOS, Linux and Docker. On Windows, the uvx launcher is typically found at %USERPROFILE%\.local\bin\uvx.exe. Confirm the path on your machine before adding it to your assistant's configuration.
It can be, provided you get it from the official GitHub repository or documentation and check the file before running it. The project is open source under the MIT license, so its code can be reviewed. Never share API keys with third-party sites, and treat analysis output as research rather than advice.
Yes. The documentation names Bitcoin (BTC) and Ethereum (ETH) among the supported coins. The tools also cover stocks, forex and futures, so coverage extends beyond crypto. Check the documentation for the symbol format your exchange expects before running a screener.
Strategies need a period of 1y or 2y so the SMA200 trend filter can finish its warmup. Shorter periods leave too few bars for that indicator to calculate. Set the period to one of those values and run the test again to see whether the results appear.
The walk-forward tool labels results as ROBUST, MODERATE, WEAK or OVERFITTED. ROBUST means the strategy held up on data it was not tuned on, while OVERFITTED suggests it fitted past noise. Neither label guarantees future results, so use the verdict as one input among many.
An MCP server is a small program that follows the Model Context Protocol, an open standard that lets AI assistants call external tools and data sources. Tradingview MCP is one such server. An assistant connected to it can request screeners or backtests during a conversation instead of relying only on its training data.
Verdict
Tradingview MCP suits people who already think in indicators and backtests and want an AI assistant to handle the repetitive legwork. It is free, open source and covers several asset classes, and its walk-forward verdict helps flag overfitted ideas early. It is less suited to complete beginners or anyone who wants a point-and-click trading interface. Start with a small test watchlist, review the outputs carefully and treat every result as a hypothesis to examine rather than a forecast.